Abstract
BACKGROUND: To avoid an invasive renal biopsy, noninvasive laboratory testing for the differential diagnosis of kidney diseases is a desirable goal. As sphingolipids are demonstrated to be involved in the pathogenesis of various kidney diseases, we investigated the possible usefulness of the simultaneous measurement of urinary sphingolipids for differentiating kidney diseases. MATERIALS AND METHODS: Residual urine specimens were collected from patients who had been clinically diagnosed with chronic glomerulonephritis (CGN), diabetic mellitus (DM), systemic lupus erythematosus (SLE), and arterial hypertension (AH). The urinary sphingolipids-CERs C16:0, C18:0, C18:1, C20:0, C22:0, and C24:0; sphingosine [Sph]; dihydrosphingosine; sphingosine 1-phosphate [S1P]; and dihydroS1P [dhS1P]-were measured by liquid chromatography-tandem mass spectrometry. Based on the results, machine learning models were constructed to differentiate the various kidney diseases. RESULTS: The urinary S1P was higher in patients with DM than in other participants (P < 0.05), whereas dhS1P was lower in the CGN and AH groups compared with control participants (P < 0.05). Sph and dhSph were higher in patients with CGN, AH, and SLE than in those with control participants (P < 0.05). The urinary CERs were significantly higher in patients with CGN, AH, and SLE than in those with control participants (P < 0.05). As a results of constructing a machine learning model discriminating kidney diseases, the resulting diagnostic accuracy and precision were improved from 94.03% and 66.96% to 96.10% and 78.26% respectively, when the urinary CERs, Sph, dhSph, S1P, dhS1P, and their ratios were added to the models. CONCLUSION: The urinary CERs, sphingoid bases, and their phosphates show alterations among kidney diseases, suggesting their potential involvement in the development of kidney injury.